Abstract: Traditional introductory computer programming practice has included writing pseudocode, code-reading and tracing, and code-writing. These problem types are often time-intensive, frustrating, cognitively complex, in opposition to learners’ self-beliefs, disengaging, and demotivating—and not much has changed in the last decade.  Pseudocode is a plain language description of the steps in a program. Code-reading and tracing involve using paper and pencil or online tools such as PythonTutor to trace the execution of a program, and code-writing requires learners to write code from scratch.  In contrast to these types of programming practice problems, mixed-up code (Parsons) problems require learners to place blocks of code in the correct order and sometimes require the correct indentation and/or selection between a distracter block and a correct code block.  Parsons problems can increase the diversity of programmers who complete introductory computer programming courses by improving the efficiency with which they acquire knowledge and the quality of knowledge acquisition itself.  This talk will feature experiments designed to investigate the problem-solving efficiency, cognitive load, pattern application and acquisition, and cognitive accessibility of adaptive Parsons problems. The results have implications for how to generate and sequence them. Bio: Carl C. Haynes-Magyar is a Presidential Postdoctoral Fellow at Carnegie Mellon University’s School of Computer Science in the Human–Computer Interaction Institute. Carl’s master’s work included evaluating curriculums based on their ability to develop a learner’s proficiencies for assessment and assessing the relationship between perceived and actual learning outcomes during web search interaction. His doctoral work involved studying the design of learning analytics dashboards (LADs) to support learners’ development of self-regulated learning (SRL) skills and investigating how people learn to program using interactive eBooks with adaptive mixed-up code (Parsons) problems. His postdoctoral work is a continued investigation into computing education that involves creating an online programming practice environment called Codespec. The goal is to scaffold the development of programming skills such as code reading and tracing, code writing, pattern comprehension, and pattern application across a gentle slope of different problem types. These types range from block-based programming problems to writing code from scratch. Codespec will support learners, instructors, and researchers by providing help-seeking features, generating multimodal learning analytics, and cultivating IDEAS: inclusion, diversity, equity, accessibility, sexual orientation and gender awareness. Carl has published several peer-reviewed articles at top venues such as the Conference on Human Factors in Computing Systems (CHI). He has taught as an instructor for courses on organizational behavior, cognitive and social psychology, human-computer interaction, learning analytics, educational data science, and data science ethics. He has been nominated for awards related to instruction and diversity, equity, and inclusion. He is a member of AAAI, ACM SIGCHI and SIGCSE, ALISE, and ISLS. Carl received his Ph.D. at the University of Michigan School of information in 2022, and a master’s degree in Library and Information Science with honors from Syracuse University’s School of Information Studies (iSchool) in 2016.